Showing posts with label Human Robot Interaction. Show all posts
Showing posts with label Human Robot Interaction. Show all posts

Anticipating Your Robot For The Best Human Robot Interaction


The Human Robot Interaction is a major challenge, but an important one, as we get more robots interacting with us in our homes and they are built with humanoid features. You see, we prefer to share our homes with things that we perceive as biological rather than unfeeling machines if we are going to have more complex helpful relationships than we do with our toaster.

We do have to be careful that we don’t make them too human-like, but don’t quite make it. Then we could end up with a creepy, zombie-like butler that would fit right in the dip in the Uncanny Valley Theory (1). Cuteness has its merits.

When we interact with another person, we anticipate and tend to copy their action as our mirror neurons fire away. That yawning spread like wildfire in that boring class we have all sat through, didn’t it? This interaction is also known as motor resonance (2). We have two things going on. We have a proactive gaze and then automatic imitation. Thus we anticipate and accommodate the motion. The result is that we feel comfortable as long as the action isn’t threatening.

Our robot butler/caregiver shouldn’t be threatening, so we have to avoid the uncanny valley with no zombies making unexpected moves. The key is to get that motor resonance functioning well when we design our humanoid helpers, but the challenge is then to find experimentally what works best with us humans in an everyday environment.

A review of methods for measuring motor resonance with the human robot interaction has been published by Scuitti et al and they show that the behavioral methodologies such as studying predictive gaze and automatic imitation measurements fit the bill much better than PET, fMRI or EEG scans.

If we can get the right human robot interaction, the advent of rent-a-robot stores may not be far away. It could work out better and more economical than have to pay for a place in a care home.

  1. http://en.wikipedia.org/wiki/Uncanny_valley
  2. A.Sciutti, A.Bisio, F.Nori, G.Metta,L.Fadiga, T.Pozzo and G.Sandini, Int. J. Soc. Robot., 223, 4, (2012)., http://www.springerlink.com/content/g82v75n437259607/fulltext.pdf

Ask The Right Question


As we get closer to having robots working with us in our everyday existence, that is, away from a production line automation situation, but in the home or the medical environment, we will have to communicate with them.

One possibility is to have a driver or better, a robot supervisor with them. But that is firstly expensive and would rule them out for most applications, and, secondly, would make the robot redundant as the person could, in many cases, perform the function that the robot would be called upon to carry out.

The consequence is that a person will have to interact with their robot without being an engineer. This means that the robot will have to ask for help and be given the information to help it do its job.

This is clearly a non-trivial problem. The parameters for success have been set out in a recent paper by Rosenthal et al in the current edition of Social Robotics (1). The team took two extant robots and with different test protocols then persuaded 48 residents of Pittsburgh to work with the robots.

The first robot, RoboSapien V2, was set to watching participants building structures out of children’s building blocks. The robot was there to record the color and shapes of the blocks used at each stage, and was allowed to ask for help with identification.

The second robot was tasked with leading the participants through a building and could enquire if it needed help with locating a particular room.

In both experiments, the participants were on time-limited tasks and could choose to ignore the robot if they wished. In fact, they tended to be more sympathetic, so took more care in answering, if the robots expressed uncertainty. The robots, though, did have to be careful with their wording and put the question in context to get a correct answer.

An interesting result was that the participants didn’t seem to be aware when the robots were uncertain although they did respond more readily. It appears that people like to be helpful, even to robots and don’t take very much trouble when asked seemingly irrelevant questions.

  1. S.Rosenthal, M.Veloso & A.K. Dey, Int, J. Soc. Robot, (2012), doi: 10.1007/s12369-012-0138-y


Website @ http://jimgoodwinbooks.com/